A three-phase network configuration type inverter zero impedance control method based on load current differential feedforward

Abstract

This invention discloses a zero-impedance control method for a three-phase grid-type inverter based on load current differential feedforward. It introduces a new differential term into the load current feedforward stage, making the inverter output impedance zero in the continuous domain, thereby significantly improving the voltage quality and disturbance rejection capability of the GFM inverter under nonlinear load conditions. The zero-impedance control method of this invention is universal and can be applied to different control structures (traditional dual-loop control or state-space control) and different reference coordinate systems (dq rotating coordinate system or αβ stationary coordinate system). A tracking differentiator (TD) based on Discrete-Time Optimal Control (DTOC) is used to perform the differential operation of the load current, ensuring that the zero-impedance control has good noise suppression capability in practical engineering environments.

Description

Technical Field

[0001] This invention relates to the field of inverters, and more particularly to a zero-impedance control method for a three-phase grid-type inverter based on load current differential feedforward. Background Technology

[0002] With the continuous expansion of renewable energy grid connection, grid-forming (GFM) inverters, as core equipment supporting voltage and frequency stability in new power systems, have attracted widespread attention for their voltage control technology. GFM inverters need to independently establish and maintain high-quality output voltage waveforms in islanded mode, and their control performance directly determines the power supply quality of the microgrid.

[0003] In existing technologies, voltage control methods for GFM inverters are mainly divided into two categories: linear methods and nonlinear methods. Among linear methods, proportional-integral (PI) controllers based on the dq rotating coordinate system and proportional-resonant (PR) controllers based on the αβ stationary coordinate system are commonly used tracking controllers. In terms of control structure, both dual-loop and single-loop control are widely adopted. In dual-loop control, the bandwidth of the outer voltage loop is constrained by the bandwidth of the inner current loop, limiting dynamic performance. While single-loop voltage controllers offer better dynamic performance, they are more sensitive to changes in LC filter parameters due to the lack of complete state feedback. In recent years, full-state feedback control methods based on state-space models have gradually developed, utilizing time-domain optimization algorithms such as pole placement and linear quadratic regulators (LQRs) to design control gains and improve dynamic performance.

[0004] In terms of disturbance rejection, load current feedforward control is an effective means of reducing inverter output impedance. Traditional load current proportional feedforward can reduce but not eliminate output impedance. Robust control and multi-resonance control can also improve disturbance rejection performance, but they have drawbacks such as difficulty in considering physical limitations, effectiveness only at specific frequency points, and large computational load. In addition, differentiators are widely used in grid-following (GFL) inverters to implement differential feedback of capacitor voltage to provide resonant damping, but their application in improving output impedance in GFM inverters is still relatively rare. The digital implementation of differentiators faces challenges such as noise amplification and phase lag. Existing differentiator solutions such as high-pass filters (HPF), lead-lag filters, and nonideal generalized integrators (nGI) are mainly optimized for the phase lag problem in the high-frequency range of GFL inverters, while the differentiating frequency band required by GFM inverters is mainly concentrated in the low-mid frequency range, making the demand for noise suppression capabilities more prominent.

[0005] The existing voltage control technology for GFM inverters mainly suffers from the following technical problems: (1) Traditional load current proportional feedforward control can only reduce but not eliminate output impedance, and the output voltage distortion is still large under nonlinear load conditions; (2) Existing output impedance shaping methods rely on specific control structures and reference coordinate systems, which are not universal enough. Furthermore, the design of state feedback parameters is coupled with the output impedance, resulting in high parameter tuning complexity. (3) Existing differentiator solutions are mainly designed for the phase lag problem of GFL inverters in the high-frequency band, and their noise suppression capability is insufficient to meet the engineering requirements of GFM inverters in the low and mid-frequency band differential applications. Summary of the Invention

[0006] This invention proposes a zero-impedance control method for a three-phase grid-type inverter based on load current differential feedforward, which can effectively solve the problems mentioned in the background art.

[0007] To achieve the above objectives, the present invention adopts the following technical solution: A zero-impedance control method for a three-phase grid-type inverter based on load current differential feedforward includes the following steps: S1. Establish the state-space model of the LC filter of the three-phase inverter; Using inductor current and capacitor voltage as state variables, inverter bridge arm midpoint voltage as control input, and load current as disturbance input, a complex state-space equation is established, applicable to dq and αβ reference coordinate systems.

[0008] S2. Construct an extended state-space model with an integrator and design a state feedback control law; Based on the traditional proportional feedforward, a differential feedforward term for the load current is added, so that the load current feedforward gain is designed as k_d = sL + k_p1 + r (dq coordinate system), so that the output impedance is strictly zero in the continuous domain; and the design of the zero impedance load current feedforward control law is completed.

[0009] S3. Derive the output impedance expression and design the zero-impedance feedforward control law; Zero impedance control completely decouples the design of the state feedback parameters (k_p1, k_p2, k_c) from the output impedance, so that the output impedance can be approximately zero regardless of how the state feedback parameters are designed.

[0010] S4. Extend zero-impedance control to different control structures and reference coordinate systems; Methods for extending zero-impedance control in different control structures and reference coordinate systems include extending it from state-space control to dual-loop control structures, from the dq coordinate system to the αβ coordinate system, and equivalent zero-impedance control structures for different inner-loop current controllers (proportional, PI, PR).

[0011] S5. Load current differentiation is achieved using a tracking differentiator based on discrete-time optimal control. The implementation method of load current differentiation based on discrete-time optimal control tracking differentiator (TD): Real-time differential estimation of load current is achieved by using the discrete recursive formula of TD. The differential accuracy and noise suppression performance are adjusted by the speed factor r and the filter factor c_0 to meet the engineering requirements of GFM inverters in low and mid-frequency differential applications.

[0012] S6, the overall implementation of the zero-impedance control system.

[0013] Combining the above steps, the state-space model, state feedback design, zero-impedance differential feedforward control law, and TD differentiator are integrated to finally construct a zero-impedance voltage control system for GFM inverters, achieving high-quality output voltage and strong disturbance rejection capability under nonlinear load conditions.

[0014] Furthermore, state feedback design methods include the LQR method, pole placement (Ackermann algorithm), or H∞ control method.

[0015] Furthermore, the tracking differentiator (TD) is replaced with a differentiator implementation scheme that has good noise suppression capability. The differentiator with good noise suppression capability must ensure differentiating accuracy in the low and mid frequency bands and noise suppression performance in the high frequency bands. The differentiator with good noise suppression capability includes an improved non-ideal generalized integrator (nGI) or an adaptive filter differentiator.

[0016] Advantages compared to existing technologies: This invention proposes a zero-impedance control method for three-phase grid-type inverters based on load current differential feedforward. By introducing a new differential term into the load current feedforward stage, the goal of achieving zero inverter output impedance in the continuous domain is realized. It has the following technical advantages: It effectively solves the problem that traditional load current proportional feedforward can only reduce but not eliminate output impedance, and significantly improves output voltage quality under nonlinear load conditions; The proposed method achieves complete decoupling between state feedback parameters and output impedance, significantly reducing the design complexity of control parameters. The proposed method is universal and applicable to different control structures (dual-loop control, state-space control) and different reference coordinate systems (dq, αβ), and does not depend on the specific controller selection. The load current differentiation operation is achieved by using a tracking differentiator (TD), which has excellent noise suppression capability and meets the engineering application requirements of GFM inverters. It maintains good performance and has satisfactory robustness even when the LC filter parameters change. Attached Figure Description

[0017] Figure 1 This is a flowchart of the zero-impedance control method for a three-phase grid inverter according to the present invention; Figure 2 This is a schematic diagram of the three-phase LC inverter topology of the present invention; Figure 3 This is a block diagram of the zero impedance control in the dq coordinate system of the present invention; Figure 4 This is a block diagram of zero-impedance control under the dual closed-loop control structure of the present invention; Figure 5 This is the equivalent zero impedance control block diagram under the dual closed-loop control structure of the present invention. Detailed Implementation

[0018] Example 1, refer to Appendix Figures 1 - 5 A zero-impedance control method for a three-phase grid-type inverter based on load current differential feedforward includes the following steps: S1. Establish the state-space model of the LC filter of the three-phase inverter; First, a state-space model of a three-phase inverter with an LC filter is established, as shown in the appendix. Figure 2 Using inductor current and capacitor voltage as state variables, inverter arm midpoint voltage as control input, and load current as disturbance input, the following state-space equations are established: dx_p / dt = A_p · x_p + B_p · u_i + E_p · w In the formula, x_p = [i_L, u_C]^T is a complex state variable, including the inductor current i_L and the capacitor voltage u_C; u_i is the midpoint voltage of the inverter bridge arm, i.e., the control input; w = i_w is the load current disturbance. Define a_1 = 1 / L, a_2 = 1 / C, where L is the filter inductor and C is the filter capacitor. The system matrix is ​​derived according to Kirchhoff's laws: A_p = [-r·a_1, -a_1; a_2, 0], B_p = [a_1; 0], E_p = [0; -a_2] In the formula, r is the equivalent series resistance of the inductor. This model is applicable to the dq rotating coordinate system and the αβ stationary coordinate system, and a resonant controller or an integral controller can be designed according to the selected reference coordinate system.

[0019] S2. Construct an extended state-space model with an integrator and design a state feedback control law; Based on the state-space model of S1, an integrator is introduced to eliminate voltage steady-state error. The state-space form of the integrator is as follows: dx_c / dt = e = u_ref - u_C In the formula, x_c represents the integrator state variable; e represents the voltage error; and u_ref represents the reference voltage. The integrator state is merged with the original system state to construct an extended state-space model. In this extended model, the integrator gain k_c can be considered as part of the extended state feedback. Then, a mature time-domain optimization algorithm is used to design the state feedback law (k_p1, k_p2, k_c). This invention employs the linear quadratic regulator (LQR) method to solve the Riccati equation to obtain the state feedback coefficients, achieving a good balance between system stability and dynamic response.

[0020] The complete expression for controlling the input is: u_i = k_r · u_ref - k_p1 · i_L - k_p2 · u_C + k_c · x_c + k_d · i_w In the formula, k_p1 and k_p2 are state feedback gains used to improve system damping; k_r is the reference feedforward gain used to improve tracking response; k_d is the load current feedforward gain used to shape output impedance; and k_c is the integrator gain used to eliminate steady-state error.

[0021] S3. Derive the output impedance expression and design the zero-impedance feedforward control law; Substituting the control inputs into the state-space model and transforming it to the s-domain, the output transfer function is derived: u_C(s) = T_r(s) · u_ref(s) - T_w(s) · i_w(s) In the formula, T_r(s) is the output-reference transfer function, and T_w(s) is the output impedance. The core objective of this invention is to design the output impedance T_w(s) to be zero in the continuous domain.

[0022] If the load current feedforward k_d remains a traditional proportional term, zero impedance cannot be achieved. Based on the analysis of the output impedance expression, designing the load current feedforward to include a differential term can achieve zero impedance (see Appendix). Figure 3 : k_d = sL + k_p1 + r (in the dq reference coordinate system) In the formula, sL is the newly added differential feedforward term; k_p1 + r is the proportional feedforward term. Compared with the traditional load current feedforward which only contains a proportional term, this invention adds a differential feedforward term sL for the load current, making the output impedance strictly zero in the continuous domain.

[0023] After zero-impedance control is implemented, the state feedback parameters (k_p1, k_p2, k_c) are completely decoupled from the output impedance, meaning that the output impedance can be approximately zero regardless of how the state feedback parameters are designed. This greatly reduces the design complexity of the state feedback parameters, and even with coarse parameter tuning, good dynamic performance and disturbance rejection can be obtained.

[0024] S4. Extend zero-impedance control to different control structures and reference coordinate systems; This invention can be extended to different control structures and reference coordinate systems.

[0025] S4.1 Generalization to the αβ stationary coordinate system: Using the Park transformation relationship k_{d,αβ}(s) = k_{d,dq}(s-jω), the zero-impedance load current feedforward gain in the αβ coordinate system is k_{d,αβ} = sL + k_p1 + r, which is consistent with the form of the dq coordinate system.

[0026] S4.2, Extended to a dual-loop control structure, see appendix. Figure 4 For the more commonly used dual-loop control (outer voltage loop, inner current loop), the output impedance is derived in the αβ stationary coordinate system. Analysis shows that zero-impedance control is independent of the voltage controller and only related to the current controller. When the inner current controller uses a proportional control coefficient k_pc, zero-impedance control can be achieved through a differentiator: K_f(s) = (sL+r) / k_pc + 1.

[0027] S4.3. Proposed Equivalent Zero Impedance Control Structure: When the inner current loop uses a PI or PR controller, the load current feedforward expression becomes complex. To address this problem, this invention proposes an equivalent zero impedance control structure, as shown in the appendix. Figure 5 This makes zero-impedance control independent of the selection of specific voltage and current controllers.

[0028] S4.4 Extension to dual-loop control in dq rotating coordinate system: When dual-loop control is implemented in dq coordinate system, the differential term sL becomes sL+jωL to achieve zero impedance in dq coordinate system.

[0029] S5. Load current differentiation is achieved using a tracking differentiator based on discrete-time optimal control. Zero-impedance control involves the differentiation of the load current. In practical systems, ideal differentiation is difficult to achieve due to noise and sampling step size limitations. Therefore, the selection and implementation of the differentiator is a key aspect of zero-impedance control.

[0030] This invention analyzes the special requirements of GFM inverters for their differentiators: the differentiating frequency band required by GFM inverters is mainly concentrated in the low harmonic frequency band of nonlinear loads (+1, -5, +7, -11, -13, etc., generally not exceeding 1 kHz), which is fundamentally different from the differentiating requirements of GFL inverters, which mainly focus on the high-frequency band near the resonant frequency. In the low-to-mid frequency band, noise suppression capability is more critical than phase lag.

[0031] Based on the above analysis, this invention employs a tracking differentiator (TD) based on discrete-time optimal control (DTOC) to achieve the differentiation of the load current. The discrete recursive formula for TD is: x_1(k + 1) = x_1(k) + h · x_2(k) x_2(k + 1) = x_2(k) + h · fhan(x_1(k) - v(k), x_2(k), r, h) In the formula, x_1(k) is the tracking signal of the input signal v(k); x_2(k) is the estimated differential signal sequence of v(k); h = c_0·T_s, where T_s is the sampling period, c_0 is the filtering factor used to achieve a balance between speed and noise suppression; r is the speed factor, which adjusts the tracking speed and accuracy; fhan is the discrete-time optimal control function.

[0032] Compared to commonly used non-ideal generalized integrators (nGI), TD exhibits almost identical frequency characteristics in the low and mid-frequency range, but with significantly smaller amplitudes and superior noise suppression capabilities in the high-frequency range. In real-world industrial environments with substantial high-frequency noise, using nGI amplifies the noise and severely impacts zero-impedance control performance, while TD effectively suppresses noise, resulting in higher quality output voltage.

[0033] S6, the overall implementation of the zero-impedance control system.

[0034] Based on the above steps, this invention constructs a complete zero-impedance control system. The specific process is as follows: First, a state-space model of the three-phase inverter LC filter is established; then, the state feedback gain is designed using the LQR method; a differential term is added to the load current feedforward stage to form a zero-impedance feedforward control law; and real-time differential calculation of the load current is implemented using a TD (Digital Transformer). In the control implementation, inductor current, capacitor voltage, and load current signals are sampled and acquired. The TD performs differential estimation of the load current, and the differential signal is superimposed with the proportional feedforward signal before being fed into the control law calculation. Finally, a modulation signal for the inverter bridge arm midpoint voltage is generated, which drives the inverter power switching transistors through pulse width modulation (PWM).

[0035] Because zero-impedance control completely decouples the state feedback parameters from the output impedance, even with roughly designed state feedback parameters using the LQR method, good voltage quality can be achieved under nonlinear load conditions. In experimental verification, the total harmonic distortion (THD) of the output voltage was reduced by nearly 40% compared to traditional control after adopting zero-impedance control; the voltage drop was significantly reduced during transient processes involving sudden linear and nonlinear loads. Furthermore, this method maintains good output voltage quality even when the LC filter parameters change, demonstrating good robustness.

[0036] Furthermore, zero impedance control can be implemented in single-phase inverter systems, and its core principle (differential feedforward of load current to make the output impedance zero) is also applicable.

[0037] Example 2: A zero-impedance control method for a three-phase grid-type inverter based on load current differential feedforward, comprising the following steps: S1. Establish the state-space model of the LC filter of the three-phase inverter; Using inductor current and capacitor voltage as state variables, inverter bridge arm midpoint voltage as control input, and load current as disturbance input, a complex state-space equation is established, applicable to dq and αβ reference coordinate systems.

[0038] The state-space equations are as follows: dx_p / dt = A_p · x_p + B_p · u_i + E_p · w In the formula, x_p = [i_L, u_C]^T is a complex state variable, including the inductor current i_L and the capacitor voltage u_C; u_i is the midpoint voltage of the inverter bridge arm, i.e., the control input; w = i_w is the load current disturbance. Define a_1 = 1 / L, a_2 = 1 / C, where L is the filter inductor and C is the filter capacitor. The system matrix is ​​derived according to Kirchhoff's laws: A_p = [-r·a_1, -a_1; a_2, 0], B_p = [a_1; 0], E_p = [0; -a_2] In the formula, r is the equivalent series resistance of the inductor. This model is applicable to the dq rotating coordinate system and the αβ stationary coordinate system, and a resonant controller or an integral controller can be designed according to the selected reference coordinate system.

[0039] S2. Construct an extended state-space model with an integrator and design a state feedback control law; Based on the traditional proportional feedforward, a differential feedforward term for the load current is added, so that the load current feedforward gain is designed as k_d = sL + k_p1 + r in the dq coordinate system, so that the output impedance is strictly zero in the continuous domain; and the design of the zero-impedance load current feedforward control law is completed.

[0040] Based on the state-space model of S1, an integrator is introduced to eliminate voltage steady-state error. The state-space form of the integrator is as follows: dx_c / dt = e = u_ref - u_C In the formula, x_c represents the integrator state variable; e represents the voltage error; and u_ref represents the reference voltage. The integrator state is merged with the original system state to construct an extended state-space model. In this extended model, the integrator gain k_c can be considered as part of the extended state feedback. Then, a mature time-domain optimization algorithm is used to design the state feedback law. This invention employs the linear quadratic regulator (LQR) method to solve the Riccati equation to obtain the state feedback coefficients, achieving a good balance between system stability and dynamic response.

[0041] The complete expression for controlling the input is: u_i = k_r · u_ref - k_p1 · i_L - k_p2 · u_C + k_c · x_c + k_d · i_w In the formula, k_p1 and k_p2 are state feedback gains used to improve system damping; k_r is the reference feedforward gain used to improve tracking response; k_d is the load current feedforward gain used to shape output impedance; and k_c is the integrator gain used to eliminate steady-state error.

[0042] S3. Derive the output impedance expression and design the zero-impedance feedforward control law; Zero-impedance control completely decouples the design of the state feedback parameters from the output impedance, ensuring that the output impedance is approximately zero regardless of how the state feedback parameters are designed. The state feedback parameters include k_p1, k_p2, and k_c.

[0043] Substituting the control inputs into the state-space model and transforming it to the s-domain, the output transfer function is derived: u_C(s) = T_r(s) · u_ref(s) - T_w(s) · i_w(s) In the formula, T_r(s) is the output-reference transfer function, and T_w(s) is the output impedance; the core objective of this invention is to design the output impedance T_w(s) to be zero in the continuous domain.

[0044] If the load current feedforward k_d is still a traditional proportional term, zero impedance cannot be achieved. Based on the analysis of the output impedance expression, the load current feedforward... in the dq reference coordinate system Designed to include differential terms, zero impedance can be achieved. k_d = sL + k_p1 + r In the formula, sL is the newly added differential feedforward term; k_p1 + r is the proportional feedforward term. Compared with the traditional load current feedforward which only contains a proportional term, this invention adds a differential feedforward term sL for the load current, making the output impedance strictly zero in the continuous domain.

[0045] After zero impedance control is implemented, the state feedback parameters are completely decoupled from the output impedance, meaning that no matter how the state feedback parameters are designed, the output impedance can be approximately zero.

[0046] S4. Extend zero-impedance control to different control structures and reference coordinate systems; Methods for extending zero-impedance control across different control structures and reference coordinate systems include extending it from state-space control to a two-loop control structure, from the dq coordinate system to the αβ coordinate system, and equivalent zero-impedance control structures for different inner-loop current controllers. Inner-loop current controllers include proportional, PI, and PR inner-loop current controllers.

[0047] The method for extending zero-impedance control in S4 to different control structures and reference coordinate systems includes the following: S4.1 Generalization to the αβ stationary coordinate system: Using the Park transformation relationship k_{d,αβ}(s) = k_{d,dq}(s-jω), the zero-impedance load current feedforward gain in the αβ coordinate system is k_{d,αβ} = sL + k_p1 + r, which is consistent with the form of the dq coordinate system.

[0048] S4.2 Extension to dual-loop control structure: For the more commonly used dual-loop control with outer loop voltage loop and inner loop current loop, the output impedance is derived in the αβ stationary coordinate system; analysis shows that zero impedance control is independent of the voltage controller and only related to the current controller; when the inner loop current controller uses a proportional control coefficient k_pc, zero impedance control can be achieved through a differentiator: K_f(s) = (sL+r) / k_pc + 1.

[0049] S4.3 Propose an equivalent zero-impedance control structure: When the inner current loop uses a PI or PR controller, the load current feedforward expression becomes complicated. To address this problem, this invention proposes an equivalent zero-impedance control structure, which makes zero-impedance control independent of the specific voltage and current controller selection.

[0050] S4.4 Extension to dual-loop control in dq rotating coordinate system: When dual-loop control is implemented in dq coordinate system, the differential term sL becomes sL+jωL to achieve zero impedance in dq coordinate system.

[0051] S5. Load current differentiation is achieved using a tracking differentiator based on discrete-time optimal control. The implementation method of load current differentiation based on discrete-time optimal control tracking differentiator (TD): Real-time differential estimation of load current is achieved by using the discrete recursive formula of TD. The differential accuracy and noise suppression performance are adjusted by the speed factor r and the filter factor c_0 to meet the engineering requirements of GFM inverters in low and mid-frequency differential applications.

[0052] The method for implementing the load current differentiation in S5 includes the following steps: The differential frequency band required by GFM inverters is mainly concentrated in the low-order harmonic frequency band of nonlinear loads, which is fundamentally different from the differential requirements of GFL inverters, which mainly focus on the high-frequency band near the resonant frequency. In the low-to-mid frequency band, noise suppression capability is more critical than phase lag. The low-order harmonic frequency band includes +1, -5, +7, -11, and -13, etc., and generally does not exceed 1kHz.

[0053] Based on the above analysis, this invention employs a tracking differentiator (TD) based on discrete-time optimal control (DTOC) to achieve the differentiation of the load current. The discrete recursive formula for TD is: x_1(k + 1) = x_1(k) + h · x_2(k) x_2(k + 1) = x_2(k) + h · fhan(x_1(k) - v(k), x_2(k), r, h) In the formula, x_1(k) is the tracking signal of the input signal v(k); x_2(k) is the estimated differential signal sequence of v(k); h = c_0·T_s, where T_s is the sampling period, c_0 is the filtering factor used to achieve a balance between speed and noise suppression; r is the speed factor, which adjusts the tracking speed and accuracy; fhan is the discrete-time optimal control function.

[0054] S6, the overall implementation of the zero-impedance control system; Combining the above steps, the state-space model, state feedback design, zero-impedance differential feedforward control law, and TD differentiator are integrated to finally construct a zero-impedance voltage control system for GFM inverters, achieving high-quality output voltage and strong disturbance rejection capability under nonlinear load conditions.

[0055] The state feedback design methods include LQR method, pole placement or H∞ control method.

[0056] Example 3, based on Example 1, replaces the LQR method for designing state feedback gain with pole placement (Ackermann algorithm) or H∞ control method.

[0057] Example 4: Based on Example 1, the tracking differentiator (TD) is replaced with a differentiator implementation scheme with good noise suppression capability. The differentiator with good noise suppression capability must ensure differentiating accuracy in the low and mid frequency bands and noise suppression performance in the high frequency bands. The differentiator with good noise suppression capability includes an improved non-ideal generalized integrator (nGI) or an adaptive filter differentiator.

Claims

1. A zero-impedance control method for a three-phase grid-type inverter based on load current differential feedforward, characterized in that: Includes the following steps: S1. Establish the state-space model of the LC filter of the three-phase inverter; Using inductor current and capacitor voltage as state variables, inverter bridge arm midpoint voltage as control input, and load current as disturbance input, a complex state-space equation is established, applicable to dq and αβ reference coordinate systems. S2. Construct an extended state-space model with an integrator and design a state feedback control law; Based on the traditional proportional feedforward, a differential feedforward term for the load current is added, so that the load current feedforward gain is designed as k_d = sL + k_p1 + r in the dq coordinate system, so that the output impedance is strictly zero in the continuous domain. Complete the design of the zero-impedance load current feedforward control law; S3. Derive the output impedance expression and design the zero-impedance feedforward control law; Zero impedance control completely decouples the design of the state feedback parameters from the output impedance. Regardless of how the state feedback parameters are designed, the output impedance can be approximately zero. The state feedback parameters include k_p1, k_p2, and k_c. S4. Extend zero-impedance control to different control structures and reference coordinate systems; Methods for extending zero-impedance control in different control structures and reference coordinate systems include extending it from state-space control to dual-loop control structures, extending it from the dq coordinate system to the αβ coordinate system, and equivalent zero-impedance control structures for different inner-loop current controllers; S5. Load current differentiation is achieved using a tracking differentiator based on discrete-time optimal control. The implementation method of load current differentiation based on discrete-time optimal control tracking differentiator: Real-time differential estimation of load current is achieved by using the discrete recursive formula of TD. The differential accuracy and noise suppression performance are adjusted by the speed factor r and the filter factor c_0 to meet the engineering requirements of GFM inverter for differential applications in low and medium frequency bands. S6, the overall implementation of the zero-impedance control system; Combining the above steps, the state-space model, state feedback design, zero-impedance differential feedforward control law, and TD differentiator are integrated to finally construct a zero-impedance voltage control system for GFM inverters, achieving high-quality output voltage and strong disturbance rejection capability under nonlinear load conditions.

2. The zero-impedance control method for a three-phase grid-type inverter based on load current differential feedforward as described in claim 1, characterized in that: The state-space equations in S1 are as follows: dx_p / dt = A_p · x_p + B_p · u_i + E_p · w In the formula, x_p = [i_L, u_C]^T is a complex state variable, including the inductor current i_L and the capacitor voltage u_C; u_i is the midpoint voltage of the inverter bridge arm, i.e., the control input; w = i_w is the load current disturbance; a_1 = 1 / L, a_2 = 1 / C are defined, where L is the filter inductor and C is the filter capacitor. The system matrix is ​​derived according to Kirchhoff's laws. A_p = [-r·a_1, -a_1; a_2, 0], B_p = [a_1; 0], E_p = [0; -a_2] In the formula, r is the equivalent series resistance of the inductor; this model is applicable to the dq rotating coordinate system and the αβ stationary coordinate system, and a resonant controller or an integral controller can be designed according to the selected reference coordinate system.

3. The zero-impedance control method for a three-phase grid-type inverter based on load current differential feedforward as described in claim 1, characterized in that: S2 includes the following steps. Based on the state-space model of S1, an integrator is introduced to eliminate voltage steady-state error; the state-space form of the integrator is as follows: dx_c / dt = e = u_ref - u_C In the formula, x_c is the integrator state variable; e is the voltage error; u_ref is the reference voltage; the integrator state is merged with the original system state to construct an extended state-space model; in this extended model, the integrator gain k_c can be regarded as part of the extended state feedback; then, the state feedback law is designed using a mature time-domain optimization algorithm. This invention uses the linear quadratic regulator method to solve the Riccati equation to obtain the state feedback coefficients, achieving a good balance between system stability and dynamic response; The complete expression for controlling the input is: u_i = k_r · u_ref - k_p1 · i_L - k_p2 · u_C + k_c · x_c + k_d · i_w In the formula, k_p1 and k_p2 are state feedback gains used to improve system damping; k_r is the reference feedforward gain used to improve tracking response; k_d is the load current feedforward gain used to shape output impedance; and k_c is the integrator gain used to eliminate steady-state error.

4. The zero-impedance control method for a three-phase grid-type inverter based on load current differential feedforward as described in claim 1, characterized in that: S3 includes the following steps. Substituting the control inputs into the state-space model and transforming it to the s-domain, the output transfer function is derived: u_C(s) = T_r(s) · u_ref(s) - T_w(s) · i_w(s) In the formula, T_r(s) is the output-reference transfer function, and T_w(s) is the output impedance; the core objective of this invention is to design the output impedance T_w(s) to be zero in the continuous domain. If the load current feedforward k_d is still a traditional proportional term, zero impedance cannot be achieved; based on the analysis of the output impedance expression, the load current feedforward... In the dq reference coordinate system Designed to include differential terms, zero impedance can be achieved. k_d = sL + k_p1 + r In the formula, sL is the newly added differential feedforward term; k_p1 + r is the proportional feedforward term; compared with the traditional load current feedforward which only contains the proportional term, the present invention adds the differential feedforward term sL of the load current, so that the output impedance is strictly zero in the continuous domain. After zero impedance control is implemented, the state feedback parameters are completely decoupled from the output impedance, meaning that no matter how the state feedback parameters are designed, the output impedance can be approximately zero.

5. The zero-impedance control method for a three-phase grid inverter based on load current differential feedforward as described in claim 1, characterized in that: The method for extending zero-impedance control in S4 to different control structures and reference coordinate systems includes the following: S4.1 Generalization to the αβ stationary coordinate system: Using the Park transformation relationship k_{d,αβ}(s) = k_{d,dq}(s-jω), the zero-impedance load current feedforward gain in the αβ coordinate system is k_{d,αβ} = sL + k_p1 + r, which is consistent with the form of the dq coordinate system; S4.2 Extension to Dual-Loop Control Structure: For the more commonly used dual-loop control with outer voltage loop and inner current loop, the output impedance is derived in the αβ stationary coordinate system. Analysis shows that zero impedance control is independent of the voltage controller and only related to the current controller. When the inner current controller uses a proportional control coefficient k_pc, zero impedance control can be achieved through a differentiator: K_f(s) = (sL+r) / k_pc + 1. S4.3 Propose an equivalent zero-impedance control structure: When the inner current loop uses a PI or PR controller, the load current feedforward expression becomes complicated; to address this problem, this invention proposes an equivalent zero-impedance control structure, so that zero-impedance control does not depend on the specific voltage controller and current controller selection. S4.4 Extension to dual-loop control in dq rotating coordinate system: When dual-loop control is implemented in dq coordinate system, the differential term sL becomes sL+jωL to achieve zero impedance in dq coordinate system.

6. The zero-impedance control method for a three-phase grid-type inverter based on load current differential feedforward as described in claim 1, characterized in that: The method for implementing the load current differentiation in S5 includes the following steps: The differential frequency band required by GFM inverters is mainly concentrated in the low harmonic frequency band of nonlinear loads, which is fundamentally different from the differential requirement of GFL inverters, which mainly focuses on the high frequency band near the resonant frequency. In the low and mid frequency bands, noise suppression capability is more critical than phase lag. Based on the above analysis, this invention employs a tracking differentiator based on discrete-time optimal control to achieve the differentiation of the load current; the discrete recursive formula for TD is: x_1(k+1) = x_1(k) + h · x_2(k) x_2(k+1) = x_2(k) + h · fhan(x_1(k)-v(k), x_2(k), r, h) In the formula, x_1(k) is the tracking signal of the input signal v(k); x_2(k) is the estimated differential signal sequence of v(k); h = c_0·T_s, where T_s is the sampling period, c_0 is the filtering factor used to achieve a balance between speed and noise suppression; r is the speed factor, which adjusts the tracking speed and accuracy; fhan is the discrete-time optimal control function.

7. The zero-impedance control method for a three-phase grid-type inverter based on load current differential feedforward as described in claim 1, characterized in that: The state feedback design methods include LQR method, pole placement or H∞ control method.

Images